Maisa: Although Artificial Intelligence remains one of the most groundbreaking sets of technologies to emerge in recent years, trust in its ability to deliver results for businesses is still being built. According to a study by The European Union, 38 per cent of people say they use scientific research and discoveries created with the help of AI, while 25 per cent still distrust it. In addition, research from Exploding Topics shows that 42 percent of companies have used AI within their company, whilst a further 40 percent are still exploring its use, demonstrating that trust has yet to be earned.

One founder looking to enhance the relationship between businesses and AI is young Spanish entrepreneur David Villalón. His company Maisa (which he co-founded with Manuel Romero) recently raised over 4.5 million euros) aims to solve reliability issues by using Artificial Intelligence by first developing a process for companies and workers to generate answers.

We caught up with the successful young founder and asked him about the concept, technology involved and just how it enables safe deployment of AI…

First of all, David, can you tell us about Maisa, in simple terms, what exactly it is, and how it works?

Maisa is an Agentic Process Automation (APA) platform designed to automate complex, knowledge-intensive business processes with full traceability. In simpler terms, we deliver cutting-edge AI reasoning and execution for commercial use. Maisa operates through a unique chain-of-work approach that ensures accountability and reliability. This is something that more traditional AI agentic solutions often fail to provide.

We offer digital workers and accountable AI agents who can easily be integrated into existing business systems. This means companies have access to autonomous analysts across supply chains, credit risk and more. Unlike traditional automation, our Digital Workers don’t just execute tasks; they continuously capture, refine, and apply organisational knowledge, improving over time through human feedback. 

Why (and when) did you decide to establish the business?

Maisa was born last year following our first-hand experience with AI models (LLMs) and Naive AI Chatbots’ RAG-based approaches. We noticed they had many limitations in business automation at scale. One of the biggest problems was the complexity and unreliability of multi-agent frameworks, resulting in hallucinations.

We realised that existing AI approaches couldn’t be trusted for complex, multi-step business processes. So, we went back to first principles and drew inspiration from operating system architecture models to build a fundamentally different AI system. That’s how Maisa and our Knowledge Processing Unit (KPU) were created.

Maisa describes itself as the first company to bring accountable AI to the market. What are the key issues surrounding the AI industry at the moment? 

I think that the AI industry is facing several major challenges. The first would be reliability. Most AI solutions today are probabilistic, meaning they can produce different results for the same input. It is this inconsistency that makes them unreliable for enterprise use. Another challenge is the lack of transparency. Most AI models are black boxes, where companies don’t know how decisions are made. This simply creates compliance risks and trust issues. Scalability is another problem, as most AI systems need to integrate with enterprise workflows, data sources, and tools. Many solutions today struggle to scale effectively across the organisation.

Another key issue is regulation and compliance. With increasing AI regulations, businesses need AI that is explainable and auditable, which most current systems are not. Lastly, complex setup and maintenance require a lot of resources and expert knowledge, so we decided to offer an all-in-one solution with rapid development and easy maintenance.

How can AI users ensure the platforms they are using are, in fact, reliable when there are so many available?

From experience, I think businesses should look for three key factors when choosing an AI platform. The first is hallucination-resistant execution because if an AI system does not produce consistent, explainable results, it’s not reliable enough for enterprise use. Auditability and compliance is another. Companies should avoid black-box AI and instead choose solutions where every AI-driven decision is traceable and logically explainable.

Scalability and maintainability also need to be considered. AI must integrate seamlessly into enterprise systems and improve over time. This will mean businesses are able to leverage progress on AI models without constant manual intervention. 

One last piece of advice I would have for AI users is not to get caught up in the latest AI models and announcements. AI models are rapidly becoming a commodity, making the choice of a model less important than the real business value the system delivers.

What advice would you have for founders who are looking to enter the AI space?

If I had to offer any advice, it would be that great founders don’t just react to the present – they anticipate what’s coming next. The most successful innovators develop a sharp intuition for shifts in technology, markets, and human behaviour, allowing them to stay ahead of the curve rather than follow it. 

In my opinion, curiosity and adaptability are powerful assets, and those who commit to deep learning can gain a significant edge. With 1,000 hours of focused effort, mastery in any field is within reach. The courage to think differently is also hugely important. You need to be able to trust your instincts and be willing to challenge conventional wisdom.

In your opinion, what are the challenges facing young entrepreneurs in tech? 

The biggest challenge is finding the balance between vision and execution. Many young entrepreneurs have big ideas yet struggle with execution in terms of turning them into viable businesses.

Another significant challenge is credibility. Breaking into industries dominated by legacy players requires young founders to prove their expertise quickly. That’s why I always recommend focusing on building a strong proof of concepts, gaining real-world experience, and surrounding yourself with the right mentors and advisors.

What are your ambitions for the company?

We want Maisa to become the global standard for AI-powered enterprise automation. Today, most AI automation solutions lack traceability and accountability. We are changing that by bringing hallucination-resistant, explainable AI to companies worldwide. We’re already making a huge impact on supply chain reliability, compliance, and decision-making for major global businesses. Over the next few years, we will expand into more high-stakes industries where AI-driven automation is critical but risky without full transparency.

If you could take any one lesson away from being a hugely successful male founder, what would it be?

I think that the biggest lesson I’ve learned is that success comes from challenging conventional wisdom. When we started Maisa, the industry was obsessed with LLMs (Language Learning Models) and RAG-based approaches despite their flaws. Instead of following that path, we went back to first principles, rethinking AI automation from the ground up. As Marc Andreessen said, “Most of the big breakthrough technologies/companies seem crazy at first… Each point in front of you is bigger than anything that ever happened.” In our case, this meant anticipating a future where AI agents interact with each other through high volumes of transactions and multiple steps, understanding that even a small error rate can cause significant problems. It was the decision to build something different that set us apart. For any founder, I think the key is to stay bold, challenge assumptions, and always be willing to rethink the way things are done.